This paper argues that empirical, conceptual, and statistical difficulties
characterise previous demonstrations that self-schemas moderate the relatio
nship between intentions and behaviour. A longitudinal study (n = 163) was
designed to overcome limitations of previous research. Theory of planned be
haviour variables, past behaviour, and self-schemas were assessed in relati
on to exercise. Behaviour was followed up two weeks later. Findings showed
that self-schemas moderated the intention-behaviour relation such that sche
matics weve more likely to enact their intentions to exercise compared to u
nschematics. Evidence suggested that the importance dimension of self-schem
a measures was responsible for the moderator effect. Self-schemas were also
associated with improved prediction of behavioural intentions after contro
lling for the other predictors. Copyright (C) 2000 John Wiley & Sons, Ltd.